The pruning algorithms for sparse least squares support vector regression machine are common methods, and easily com- prehensible, but the computational burden in the training phase is heavy due to the retraining in p...The pruning algorithms for sparse least squares support vector regression machine are common methods, and easily com- prehensible, but the computational burden in the training phase is heavy due to the retraining in performing the pruning process, which is not favorable for their applications. To this end, an im- proved scheme is proposed to accelerate sparse least squares support vector regression machine. A major advantage of this new scheme is based on the iterative methodology, which uses the previous training results instead of retraining, and its feasibility is strictly verified theoretically. Finally, experiments on bench- mark data sets corroborate a significant saving of the training time with the same number of support vectors and predictive accuracy compared with the original pruning algorithms, and this speedup scheme is also extended to classification problem.展开更多
An intuitive method for circle fitting is proposed. Assuming an approximate circle(CA,n) for the fitting of some scattered points, it can be imagined that every point would apply a force to CA,n, which all together fo...An intuitive method for circle fitting is proposed. Assuming an approximate circle(CA,n) for the fitting of some scattered points, it can be imagined that every point would apply a force to CA,n, which all together form an overall effect that "draws" CA,n towards best fitting to the group of points. The basic element of the force is called circular attracting factor(CAF) which is defined as a real scalar in a radial direction of CA,n. An iterative algorithm based on this idea is proposed, and the convergence and accuracy are analyzed. The algorithm converges uniformly which is proved by the analysis of Lyapunov function, and the accuracy of the algorithm is in accord with that of geometric least squares of circle fitting. The algorithm is adopted to circle detection in grayscale images, in which the transferring to binary images is not required, and thus the algorithm is less sensitive to lightening and background noise. The main point for the adaption is the calculation of CAF which is extended in radial directions of CA,n for the whole image. All pixels would apply forces to CA,n, and the overall effect of forces would be equivalent to a force from the centroid of pixels to CA,n. The forces from would-be edge pixels would overweigh that from noisy pixels, so the following approximate circle would be of better fitting. To reduce the amount of calculation, pixels are only used in an annular area including the boundary of CA,n just in between for the calculation of CAF. Examples are given, showing the process of circle fitting of scattered points around a circle from an initial assuming circle, comparing the fitting results for scattered points from some related literature, applying the method proposed for circular edge detection in grayscale images with noise, and/or with only partial arc of a circle, and for circle detection in BGA inspection.展开更多
锂电池荷电状态(state of charge,SOC)的准确估计依赖于精确的锂电池模型参数。在采用带遗忘因子的递推最小二乘法(forgetting factor recursive least square,FFRLS)对锂电池等效电路模型进行参数辨识时,迭代初始值选取不当会造成辨识...锂电池荷电状态(state of charge,SOC)的准确估计依赖于精确的锂电池模型参数。在采用带遗忘因子的递推最小二乘法(forgetting factor recursive least square,FFRLS)对锂电池等效电路模型进行参数辨识时,迭代初始值选取不当会造成辨识精度低、收敛速度慢的问题。为此,将电路分析法与FFRLS相结合,提出基于改进初值带遗忘因子的递推最小二乘法(improved initial value-FFRLS,IIV-FFRLS)。首先,通过离线辨识得到各荷电状态点对应的等效电路模型参数并进行多项式拟合;然后,利用初始开路电压(open circuit voltage,OCV)和OCV-SOC曲线获得初始SOC,代入参数拟合函数得到初始参数;最后,将初始参数带入递推公式得到IIV-FFRLS迭代初始值。对4种锂电池工况进行参数辨识,结果表明:与传统方法相比,IIV-FFRLS的平均相对误差、收敛时间分别减小58%、23%以上;IIV-FFRLS具有更高的辨识精度与更快的收敛速度。展开更多
基于点云的空间非合作目标位姿估计,常受到噪声影响.提出截断最小二乘估计与半定松弛(truncated least squares estimation and semidefinite relaxation,TEASER)与迭代最近点(iterative closest point,ICP)的结合算法,提升空间非合作...基于点云的空间非合作目标位姿估计,常受到噪声影响.提出截断最小二乘估计与半定松弛(truncated least squares estimation and semidefinite relaxation,TEASER)与迭代最近点(iterative closest point,ICP)的结合算法,提升空间非合作目标位姿估计精度与鲁棒性.该方法包括粗配准与精配准两个环节:在粗配准环节中,基于局部点云与模型点云的方向直方图特征(signature of histogram of orientation,SHOT)确定匹配对,利用TEASER算法求解初始位姿;在精配准环节中,可结合ICP算法优化位姿估计结果.北斗卫星仿真实验表明:在连续帧位姿估计中,噪声标准差为3倍点云分辨率时,基于TEASER的周期关键帧配准方法的平移误差小于3.33 cm,旋转误差小于2.18°;与传统ICP方法相比,平均平移误差与平均旋转误差均有所降低.这表明所提出的空间非合作目标位姿估计方法具有良好的精度和鲁棒性.展开更多
基金supported by the National Natural Science Foundation of China(50576033)
文摘The pruning algorithms for sparse least squares support vector regression machine are common methods, and easily com- prehensible, but the computational burden in the training phase is heavy due to the retraining in performing the pruning process, which is not favorable for their applications. To this end, an im- proved scheme is proposed to accelerate sparse least squares support vector regression machine. A major advantage of this new scheme is based on the iterative methodology, which uses the previous training results instead of retraining, and its feasibility is strictly verified theoretically. Finally, experiments on bench- mark data sets corroborate a significant saving of the training time with the same number of support vectors and predictive accuracy compared with the original pruning algorithms, and this speedup scheme is also extended to classification problem.
基金Project(2013CB035504) supported by the National Basic Research Program of ChinaProject(2012zzts078) supported by the Fundamental Research Funds for the Central Universities of Central South University,ChinaProject(2009ZX02038) supported by the National Science and Technology Major Project of the Ministry of Science and Technology of China
文摘An intuitive method for circle fitting is proposed. Assuming an approximate circle(CA,n) for the fitting of some scattered points, it can be imagined that every point would apply a force to CA,n, which all together form an overall effect that "draws" CA,n towards best fitting to the group of points. The basic element of the force is called circular attracting factor(CAF) which is defined as a real scalar in a radial direction of CA,n. An iterative algorithm based on this idea is proposed, and the convergence and accuracy are analyzed. The algorithm converges uniformly which is proved by the analysis of Lyapunov function, and the accuracy of the algorithm is in accord with that of geometric least squares of circle fitting. The algorithm is adopted to circle detection in grayscale images, in which the transferring to binary images is not required, and thus the algorithm is less sensitive to lightening and background noise. The main point for the adaption is the calculation of CAF which is extended in radial directions of CA,n for the whole image. All pixels would apply forces to CA,n, and the overall effect of forces would be equivalent to a force from the centroid of pixels to CA,n. The forces from would-be edge pixels would overweigh that from noisy pixels, so the following approximate circle would be of better fitting. To reduce the amount of calculation, pixels are only used in an annular area including the boundary of CA,n just in between for the calculation of CAF. Examples are given, showing the process of circle fitting of scattered points around a circle from an initial assuming circle, comparing the fitting results for scattered points from some related literature, applying the method proposed for circular edge detection in grayscale images with noise, and/or with only partial arc of a circle, and for circle detection in BGA inspection.
文摘基于点云的空间非合作目标位姿估计,常受到噪声影响.提出截断最小二乘估计与半定松弛(truncated least squares estimation and semidefinite relaxation,TEASER)与迭代最近点(iterative closest point,ICP)的结合算法,提升空间非合作目标位姿估计精度与鲁棒性.该方法包括粗配准与精配准两个环节:在粗配准环节中,基于局部点云与模型点云的方向直方图特征(signature of histogram of orientation,SHOT)确定匹配对,利用TEASER算法求解初始位姿;在精配准环节中,可结合ICP算法优化位姿估计结果.北斗卫星仿真实验表明:在连续帧位姿估计中,噪声标准差为3倍点云分辨率时,基于TEASER的周期关键帧配准方法的平移误差小于3.33 cm,旋转误差小于2.18°;与传统ICP方法相比,平均平移误差与平均旋转误差均有所降低.这表明所提出的空间非合作目标位姿估计方法具有良好的精度和鲁棒性.
文摘联合对角化方法是求解盲源分离问题的有力工具.但是现存的联合对角化算法大都只能求解实数域盲源分离问题,且对目标矩阵有诸多限制.为了求解更具一般性的复数域盲源分离问题,提出了一种基于结构特点的联合对角化(Structural Traits Based Joint Diagonalization,STBJD)算法,既取消了预白化操作解除了对目标矩阵的正定性限制,又允许目标矩阵组为复值,具有极广的适用性.首先,引入矩阵变换,将待联合对角化的复数域目标矩阵组转化为新的具有鲜明结构特点的实对称目标矩阵组.随后,构建联合对角化最小二乘代价函数,引入交替最小二乘迭代算法求解代价函数,并在优化过程中充分挖掘所涉参量的结构特点加以利用.最终,求得混迭矩阵的估计并据此恢复源信号.仿真实验证明与现存的有代表性的对目标矩阵无特殊限制的复数域联合对角化算法FAJD算法及CVFFDIAG算法相比,STBJD算法具有更高的收敛精度,能有效地解决盲源分离问题.
